Back

Distinct neural geometries for target position and velocity in the primate superior colliculus

Yang, F.; Bourrelly, C.; Eldridge, M.; Gandhi, N. J.

2026-06-26 neuroscience
10.64898/2026.06.22.733795 bioRxiv
Show abstract

Moving targets pose a fundamental challenge for sensorimotor control: by the time a targets position is transformed into neural signals, the target has already moved. Primates compensate for this delay and accurately intercept objects by accounting for target velocity. However, the neural mechanisms by which sensorimotor structures represent target velocity remain poorly understood. Here we recorded spikes and local field potentials from the superior colliculus (SC) of male macaques viewing or directing saccades to stationary and moving targets and used time-resolved dimensionality reduction and classification to characterize SC population dynamics. Target velocity emerged rapidly within the visual response and was robustly decodable despite weak and inconsistent single-neuron motion sensitivity. In low-dimensional state space, activity formed a V-shaped manifold in which speed was organized continuously along direction-specific arms. This geometry generalized across animals and signal modalities, could not be reproduced by displacement-based simulations, and remained largely dissociable from the axis encoding target location. Velocity information was strongest near the border between superficial and intermediate layers, identifying a laminar locus where visual motion signals may be transformed into commands for interception. These findings reveal separate, partially overlapping geometries for representation of target speed and position in the SC and overturn the longstanding view that it encodes only target location and movement goals. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=180 SRC="FIGDIR/small/733795v1_ufig1.gif" ALT="Figure 1"> View larger version (37K): org.highwire.dtl.DTLVardef@473955org.highwire.dtl.DTLVardef@52399org.highwire.dtl.DTLVardef@164254forg.highwire.dtl.DTLVardef@1cd7395_HPS_FORMAT_FIGEXP M_FIG C_FIG

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.